Technical Product Manager

About the role

Bobyard is transforming the construction sector by replacing traditional, error-prone manual cost estimation workflows with advanced computer vision and natural language processing technologies. The company seeks a technical leader to take ownership of model-driven product surfaces, functioning as a direct partner to engineering teams to resolve intricate performance and user experience challenges. This position demands a practitioner who can bridge the gap between complex machine learning systems and tangible product value for construction professionals. You will operate at the intersection of model capability and user need, ensuring that probabilistic outputs translate into reliable, actionable features. The role requires a high tolerance for ambiguity and a rigorous approach to validating model behavior in production environments.

Key facts

What you'll do

• Own the complete lifecycle of features powered by machine learning models, steering them from initial customer discovery and problem definition through to final deployment and post-launch monitoring.
• Interpret confusion matrices, precision-recall curves, and detailed evaluation reports to engage in substantive technical debates with computer vision and ML engineers regarding model architecture and data strategy.
• Navigate and articulate trade-offs between inference latency, computational cost, and prediction accuracy, communicating the business implications of these technical decisions to executive stakeholders and clients.
• Design, curate, and maintain rigorous evaluation datasets and benchmarks to enable objective, reproducible measurement of model performance across diverse construction document types.
• Assess production readiness by analyzing the downstream consequences of specific failure modes, such as hallucinated quantities or misclassified trade categories, on the estimation workflow.
• Synthesize ambiguous or qualitative user feedback from estimators and project managers into precise technical specifications, prompt engineering adjustments, or user interface refinements.
• Function as a technical peer to ML and CV engineers during daily collaboration, contributing to architectural discussions and data labeling strategies rather than acting solely as a project coordinator.
• Define and track key product metrics that correlate model performance improvements with user efficiency gains and estimation accuracy uplift.
• Coordinate cross-functionally with sales and customer success to scope pilot programs and manage expectations regarding model capabilities during pre-sales engagements.
• Establish and enforce data quality standards for training and evaluation sets, working closely with annotation teams to ensure label consistency for difficult edge cases in construction plans.
• Drive the product roadmap for the core estimation engine, prioritizing model improvements and feature work based on ROI analysis and strategic differentiation.
• Conduct competitive analysis on emerging foundation models and vendor APIs to evaluate build-versus-buy decisions for specific perception tasks.

Requirements

• Minimum of four years of product management experience with direct, hands-on ownership of machine learning or artificial intelligence features shipped to production.
• Deep, practical understanding of model evaluation methodologies, including the nuances of precision, recall, F1 scores, IoU, and the business trade-offs inherent in threshold tuning for imbalanced datasets.
• Demonstrated history of successfully launching zero-to-one products or executing major, high-risk iterations within a B2B SaaS context, preferably in vertical software or regulated industries.
• Ability to serve as the definitive authority on model viability and risk assessment for both C-suite leadership and external enterprise clients during high-stakes negotiations.
• Demonstrated capacity to rapidly acquire deep domain expertise in the construction industry - specifically estimating workflows, CSI codes, and plan reading - within the first thirty days.
• Strong bias for action and a documented preference for iterative, transparent development cycles over heavy bureaucratic processes or waterfall planning.
• Exceptional written and verbal communication skills, capable of translating complex probabilistic concepts for non-technical audiences without loss of fidelity.
• Proven ability to manage stakeholder expectations when model performance plateaus, pivoting strategy toward human-in-the-loop workflows or product pivots as necessary.

Nice to have

• Direct professional experience managing products built on computer vision systems, specifically object detection, segmentation, or OCR pipelines applied to unstructured documents.
• Previous tenure at companies operating in the AI infrastructure, geospatial analytics, or construction technology space, such as Scale AI, Labelbox, Samsara, Matterport, Cape Analytics, or DroneDeploy.
• Proficiency in Figma for high-fidelity prototyping and design system management, coupled with the ability to read and comprehend Python, PyTorch, or TensorFlow codebases to facilitate deeper engineering collaboration.
• Familiarity with MLOps tooling for experiment tracking, model registry, and automated retraining pipelines (e.g., MLflow, Weights & Biases, Kubeflow).
• Experience with retrieval-augmented generation (RAG) architectures or large language model fine-tuning for domain-specific knowledge extraction.

Skills & tools

• Computer Vision
• Natural Language Processing (NLP)
• Model Evaluation & Metrics (Precision, Recall, mAP, F1)
• Product Roadmap Management & Prioritization Frameworks
• B2B SaaS Product Development
• Figma
• Python (Code Literacy)
• MLOps / Experiment Tracking
• Construction Technology (ConTech) Domain Knowledge
• Stakeholder Management & Technical Communication

Practical notes

This position requires a full-time, on-site presence in San Francisco; remote arrangements are not available for this role. The organization prioritizes candidates who are prepared to collaborate in person to accelerate the development velocity and culture of the early-stage team. Visa sponsorship details are not explicitly stated in the listing; candidates requiring work authorization should inquire directly during the initial screening phase. The interview process typically involves a technical deep-dive with the engineering team, a product case study focused on ML evaluation, and conversations with leadership.